Using a delay time T=5.0 gives a reconstruction that appears more complicated and less faithful to the original phase space:
.
On the other hand taking "too small" a delay (e.g. T=.05) gives the following reconstruction:
.
Clearly in this case the basic correlation is that X(t+T) is nearly equal to X(t), and the reconstruction relies on the small deviations from this.
Return to text